Hi Kevin, WLM is indeed a complex beast. I will not pretend to be able to completely answer all of your questions here, as there are a number of variables involved, but I will attempt to provide some insight.
Firstly, there are two algorithms that WLM uses to decide on the assignment of Dynamic PAV's. The first is the GOAL algorithm. This is the one we think of when WLM is trying to 'help' a workload. In order to come into play, you do not just need IOSQ time, you need a Service Class Period that is not meeting its goal due to I/O queue time and a free PAV from either an idle disk or another Service Class Period that can afford to donate. WLM will weigh the pros and cons of the move before making it. Note that this action only takes place every 30 seconds. The second is the Efficiency algorithm. In this case, WLM will survey all of the I/O activity and attempt to distribute the available PAV's to provide the best overall throughput. This algorithm runs every 60 seconds. For a detailed understanding of what WLM is thinking, you need to collect some type 99 SMF records. As you may know, they are quite lengthy and are likely turned off in your shop. Perhaps try temporarily turning them on during a period when you would expect to see movement (15 min or so should give you some reasonable insight). Finally, while this is not the forum for a sales pitch, talk to us at IntelliMagic. We can help. Best regards, John Baker ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
